381

Antibodies to Recombinant-Derived Glycoprotein B after Natural Human Cytomegalovirus Infection Correlate with Neutralizing Activity Gary S. Marshall, Gerard P. Rabalais, Gordon G. Stout, and Sharon L. Waldeyer

University ofLouisville School ofMedicine. Louisville. Kentucky

The rationale for vaccine immunoprophylaxis against congenital and acquired human cytomegalovirus (HCMV) infection is based on experimental data, clinical observations, and the direct demonstration of protective efficacy with a live attenuated HCMV strain [1]. However, concerns about the potential for oncogenicity, latency, reactivation, and induction of chronic disease have prompted many investigators to search for inert virion subunits that mimic the immunogenic capacity of live whole virus. One important candidate subunit vaccine immunogen is the abundant envelope glycoprotein complex, glycoprotein B (gB), a disulfidelinked complex with components of 130-116 and 58-55 kDa. Monoclonal antibodies that recognize gB can neutralize HCMV [2], and immunoaffinity-purified gB or gB produced by recombinant virus vectors can induce neutralizing antibody and cellular responses in experimental animals [35]. Recently, human inoculations with immunoaffinity-purified gB resulted in neutralizing antibodies and lymphocyte proliferative responses in previously seronegative individuals and booster responses in naturally immune volunteers [6]. Despite the above observations, studying the immune response to gB after natural infection has been problematic because of the large number of HCMV-associated infectedcell proteins, which make interpretation of immunoblot and

Received 16 July 1991; revised 20 September 1991. Presented in part: 30th Interscicnce Conference on Antimicrobial Agents and Chemotherapy. Octobcr 1990. Atlanta (abstract 205); XVI International Herpesvirus Workshop, July 1991, Pacific Grove, California (abstract 4A-17). Financial support: Alliant Community Trust Fund (89-07 and 90-14), Alliant Hcalth Systems, Louisville, Kentucky. Reprints or correspondence: Dr. Gary S. Marshall. Department ofPediatrics, University of Louisville School of Medicine. Louisville, KY 40292. The Journal of Infectious Diseases 1992;165:381-4 © 1992 by The University of Chicago. All rights reserved. 0022-1899/92/6502-0027$01.00

immunoprecipitation assays difficult. Assays using virions or infected cells as antigen consistently show reactive proteins in the appropriate molecular mass ranges, but these can be obscured by other antigenic proteins [7]; furthermore, molecular mass determinations can vary between laboratories, making it difficult to compare studies from different institutions. Liu et al. [8], using gB partially purified by anion-exchange high-pressure liquid chromatography, were able to demonstrate consistent humoral responses in a small group of HCMV-immune individuals. More recently, Britt et al. [9] used a vaccinia-gB recombinant to study gB antibody in convalescent human sera. In the present study, an adenovirus recombinant that expresses HCMV gB (Ad-gB) was used to study the humoral immune response in 48 seropositive individuals by radioimmunoprecipitation (RIP) and specific absorption.

Materials and Methods Cells. viruses. and other reagents. The construction of Ad-gB has been described previously [5]. Briefly, the gene-encoding HCMV Towne strain gB was cloned into adenovirus type 5 (Ad5) downstream from the E3 promoter, deleting nonessential E3 sequences. Peak levels of immunoreactive gB protein were produced by Ad-gB 24-36 h after synchronous infection of A549 (human lung carcinoma) cells, and the recombinant virus induced HCMV-neutralizing and polypeptide-specific antibodies in hamsters after intranasal inoculation. Ad-gB and the parental Ad-5 strain were routinely propagated on A549 cells grown in Eagle's MEM containing 10% fetal bovine serum. Anti-gB monoclonal antibody 15D8 [2] was a gift of L. Rasmussen (Stanford University, Stanford, CA), and a polyclonal, monospecific gB antiserum raised in guinea pigs [3] was provided by E. G6ncz61 (Wistar Institute, Philadelphia, PA). Antigen preparations. Uninfected cells or cells infected with Ad-5 or Ad-gB were labeled 24-48 h after infection with 10 J.LCi/ml [35 S]methionine in methionine-free medium containing

Downloaded from http://jid.oxfordjournals.org/ at University of Iowa Libraries/Serials Acquisitions on June 25, 2015

Glycoprotein B (gB) of human cytomegalovirus (HCMV) was partially purified by lentil-lectin column chromatography from cells infected with an adenovirus-gB recombinant. This antigen, which contained specifically reactive proteins of ,..,.130 and 55 kDa, was used to investigate gB antibody levels after natural HCMV infection in 48 individuals. All sera had IgG antibody to gB as detected by radioimmunoprecipitation (RIP) assays. Quantitative RIP showed a strong correlation between gB antibody and neutralizing activity (r = .74, P < .001) but a weak correlation between gB antibody and total HCMV-specific IgG (r = .36, P < .02). When gB antibody was specifically absorbed from 20 serum specimens, neutralizing antibody titer was reduced a median of 48% (range, 0-98%). These data confirmed that antibodies to gB are a large component of the neutralizing antibody response to HCMV and support a role for this protein in the development of subunit vaccines.

382

Concise Communications

in Ad-gB lanes and the corresponding regions from adjacent control (Ad-5) lanes were excised and counted in a scintillation counter. gB-specific cpm were defined as gB cpm minus the background cpm from corresponding regions of control lanes. Absorptions. An insoluble precipitate of Ad-5- or Ad-gBinfected A549 cells was made by acetone extraction as described. At 4 days after infection, when maximum cytopathic effect was present, cells were harvested, washed, and resuspended in PBS. After addition of 4 vol of acetone and incubation for 30 min at 4°C, the precipitate was collected by centrifugation, dried, resuspended in PBS containing 1% bovine serum albumin, and frozen. Preliminary titration experiments indicated that absorption with 7.5 mg (dry weight) of Ad-gB-infected cell acetone precipitate was enough to maximally reduce neutralizing activity in three different serum samples; this amount of precipitate corresponded to 107 cells, identical to the number of recombinant-infected cells used for absorption in another study [9]. Accordingly, 20 serum specimens were absorbed overnight at 4°C with this amount of Ad-5- or Ad-gBderived precipitate; after absorption, the precipitate was pelleted by centrifugation, and the supernatants were tested for neutralizing activity. Study sera. Anonymously coded sera from 48 HCMV antibody-positive individuals were used in this study. Specimens were obtained from the Clinical Immunology Laboratory of the Alliant Health System, the University of Louisville Tissue Typing Laboratory, and research laboratory personnel. Additional specimens from normal children were provided by S. Starr (Children's Hospital of Philadelphia). In alL the population consisted of 23 patients with chronic renal disease awaiting transplant, 16 normal children, 8 healthy adults, and 1 child after liver transplant. Patients known to be actively infected with HCMV were excluded from study.

Results Characterization of recombinant-derived antigen. Lectinpurified glycoproteins from Ad-gB-infected cells contained two major bands that were specifically reactive with gB antisera in RIP assays. As shown in figure I (lanes 3), monoclonal antibody 1508, monospecific guinea pig antiserum, and convalescent human serum all recognized a high-molecularmass protein (126-122 kOa) and a smaller moiety (60-57 kOa). These proteins were not present in lectin-purified AdS-infected cell glycoprotein preparations (lanes 2) or in identically prepared antigen from uninfected cells (lanes I). The upper and lower gB bands probably represent (respectively) the glycosylated precursor protein gp 130 and the carboxy-terminal cleavage product, gp55 [10]. Additional bands in lanes 2 and 3 could represent adenoviral or adenovirus-induced cellular glycoproteins that copurify with gB and are specifically or nonspecifically bound by antibody. Some of these proteins may have remained associated with lentil lectin, which would bind immunoglobulins by virtue of their glysosidic residues. Specific bands in the range of 93 kOa, corresponding to the amino terminal cleavage product gp93

Downloaded from http://jid.oxfordjournals.org/ at University of Iowa Libraries/Serials Acquisitions on June 25, 2015

I% fetal bovine serum. Immediately after labeling, cells were harvested, washed, and lysed by sonication in buffer containing ISO mM NaCl, 20 mM TRIS (pH 7.5), 1% Nonidet-P40, 1% deoxycholate, and 0.1 mM phenyl methylsulfonyl fluoride. The lysate was clarified by centrifugation at 140,000 g for I h at 4°C, then passed over lentil-lectin Sepharose 4B (Sigma, St. Louis) columns equilibrated with lysis buffer at room temperature. Columns were then washed with 10-bed volumes of running buffer containing ISO mM NaCl, 20 mM TRIS (pH 7.5), 0.1 % Nonidet-P40, and 0.1 %deoxycholate, and bound glycoproteins were eluted with 6-bed volumes of running buffer containing 0.2 M methyl a-D-mannopyranoside. Eluted glycoproteins were concentrated by ultrafiltration in centrifugal concentrators (10,000 molecular weight cutoff; Amicon, Danvers, MA), and the specific activity of the final concentrate was determined; 5 X 104 counts per minute (cpm) of antigen was used in each reaction with antibody. All reactions used the same lot of antigen and were done and analyzed within a 3-week period. Immunologic assays. Neutralizing antibody was measured by an ELISA-based color reduction method (unpublished data). Briefly, a standard (Towne strain) HCMV inoculum was incubated for 1 h with serial dilutions of serum and centrifuged onto foreskin fibroblast monolayers growing in 96-well plates. At 40 h after infection, monolayers were fixed in situ with cold ethanol/acetone and reacted sequentially with a monoclonal antibody to the 72-kDa immediate-early HCMV protein (Chemicon, Temecula, CA) and peroxidase-conjugated goat antimouse immunoglobulin (Accurate, Westbury, NY). After addition of 2,2'-azinobis(3-ethylbenzthiazoline-6-sulfonic acid) and H 2 0 2 , absorbance was determined spectrophotometrically. The neutralizing titer was defined as the interpolated serum dilution resulting in 50% reduction in absorbance compared with control wells receiving inoculum without serum (this titer was shown to correlate well with plaque reduction assay). Total HCMV-specific IgG was determined by a commercially available ELISA kit (Cytomegelisa; Whittaker, Walkersville, MD) that uses strain AD 169 on the solid phase. For RIP, antigens were preabsorbed for 1 h at 4°C with pooled seronegative serum, and resultant antigen-antibody complexes were removed with protein G-agarose beads (ImmuBind; Genex Corp., Gaithersburg, MD). Preabsorbed antigen was then combined with antibody in buffer consisting of 10 mM sodium phosphate, 150 mM NaCl, 1% Triton X-IOO, and 10 mM EDTA (pH 7.0); final volume of the reaction mixture was 100 JLl and final serum dilution was 1: 50. After incubation at 4°C overnight, protein G beads were added and the mixture was agitated for 1 h at 4°C. Beads were then collected by centrifugation and washed five times in cold buffer. Bound antigen-antibody complexes were eluted from the beads by heating at 90°C for 10 min in 100 mM TRIS (pH 6.8), 2% SDS, 40% glycerol, 0.1 %Triton X-I 00, and 10% 2-mercaptoethanol, and this supernatant was combined with an equal volume of 0.01 % bromophenol blue before electrophoresis through reducing 7.5% polyacrylamide gels. Gels were treated with Coomassie brilliant blue and destained overnight with methanol/acetic acid to visualize immunoglobulin bands (which served as lane markers) and to fix the separated proteins. Gels were then treated with scintillant (Resolution; E. M. Corp., Chestnut Hill, MA), dried, and autoradiographed. To quantitate gB-specific antibody levels, gB bands

JID 1992;165 (February)

Concise Communications

JID 1992; 165 (February)

2

3

2

3

2

3

2

3

kDa

200 -

97 -

46 -

overnight with acetone precipitates derived from either Ad5-infected (mock absorbed) or Ad-gB-infected (gB absorbed) cells. The neutralizing titer of each serum specimen after mock absorption was determined, and the neutralizing activity remaining at that dilution after gB absorption was then measured. The percentage reduction in neutralizing activity was expressed as 100 X (% neutralization mock absorbed - % neutralization gB absorbed)/% neutralization mock absorbed. After absorption, decreases in gB antibody detected by RIP were documented (not shown). For all 20 sera, the median reduction in neutralizing activity after absorption was 48% (range, 0-98%). Some reduction in neutralizing activity was seen in 18 of 20 sera, whereas gB absorption did not change the neutralizing activity of 2 specimens. When the 4 sera exhibiting 80% reduction, there were no differences in median neutralizing titer or gB-specific cpm before absorption.

Discussion 30 -

Figure 1. Radioimmunoprecipitation assay. Each antiserum (listed at top) was reacted with 5 X 104 cpm of three antigens (lanes 1-3); immunoprecipitated proteins were electrophoresed through polyacrylamide gels and autoradiographed for 3 days. Antisera: 15D8, glycoprotein B (gB)-specific monoclonal antibody (I :20); GP, gB-specific guinea pig antiserum (I :50); POS, human cytomegalovirus (HCMV) antibody-positive human serum (1:50); NEG, HCMV antibody-negative human serum (I :50). Antigen preparations consisted of 35S-labeled lectin-purified glycoproteins from uninfected A549 cells (lanes I), adenovirus type 5-infected A549 cells (lanes 2), or A549 cells infected with adenovirus recombinant expressing HCMV gB (lanes 3). Molecular mass markers are shown at left; arrows indicate upper (gp 130) and lower (gp55) gB bands.

These data provided strong evidence for the role ofgB-specific antibody in serum neutralizing activity against HCMV. The study population was selected to represent the "steady state" of HCMV antibody after natural infection in healthy individuals or those with chronic disease; attempts were made to exclude severely immunocompromised patients and those with active infection. The availability of Ad-gB provided the opportunity to examine responses to this protein in 4

a •

Neutralizing Ab Total CMV-IgG

3

a

a

a

[10], were not detected in any assays using monoclonal antibody, guinea pig antiserum, or human serum. Correlation between gB antibody and neutralizing activity. All 48 known HCMV antibody-positive human sera that were tested contained IgG antibody to gB. As seen in figure 2, the correlation between gB antibody, represented as log gB-specific cpm, and neutralizing activity, represented as log neutralizing antibody titer, was strong (Spearman rank, r = .74; significance by t distribution, P < .00 I). The sera were also tested for total HCMV-specific IgG by ELISA, represented as optical density in figure 2. The correlation between gB antibody and HCMV IgG was poor (r = .36; P < .02), suggesting that the presence ofgB antibody was not simply a marker for HCMV IgG. Specific absorptions. To analyze further the relationship suggested above between neutralizing activity and gB antibody level, 20 of the seropositive specimens were absorbed

2

•

a

rI'

I will

••

.).

••

r·· ·

•

o~:"'-_"""'--"""T'"----'-----r-----'-----'

o

2

3

Log gB-8peclflc CPU Figure 2. Correlation between glycoprotein B (gB) antibody (expressed as log gB-specific counts per minute [CPM]) and neutralizing antibody (Ab) titer (log reciprocal titer) or total human cytomegalovirus (CMV) IgG measured by ELISA (optical density, OD). r = Spearman rank correlation coefficient.

Downloaded from http://jid.oxfordjournals.org/ at University of Iowa Libraries/Serials Acquisitions on June 25, 2015

69 -

383

384

Concise Communications

are data to suggest that humoral immunity can also be protective. It is likely that neutralizing antibodies are the most biologically relevant class of antibody in this regard. This study confirms that antibody responses to gB are an important component of the neutralizing antibody response to HCMV and supports efforts to use this molecule as a subunit vaccine component. Acknowledgments

We thank S. Plotkin, S. Starr, C. Harrison, and L. Hunt for helpful suggestions and K. Cost and J. Hazell for assistance in specimen aquisition. References I. Marshall GS. Plotkin SA. Progress toward developing a cytomegalovirus vaccine. Infect Dis Clin North Am 1990;4:283-98. 2. Rasmussen L, Mullenax J, Nelson R, Merigan TC Viral polypeptides detected by a complement-dependent neutralizing murine monoclonal antibody to human cytomegalovirus. J Virol 1985;55:274-80. 3. G6ncz61 E, Hudecz F, Ianacone J, Dietzschold B, Starr S, Plotkin SA. Immune responses to isolated cytomegalovirus envelope proteins. J Virol 1986;58:661-4. 4. Cranage MP, Kouzarides T, Bankier AT, et al. Identification of the human cytomegalovirus glycoprotein B gene and induction of neutralizing antibodies via its expression in recombinant vaccinia virus. EMBO J 1986;5:3057-63. 5. Marshall GS, Ricciardi RP, Rando RF, et al. An adenovirus recombinant that expresses the human cytomegalovirus major envelope glycoprotein and induces neutralizing antibodies. J Infect Dis 1990; 162: 1177-81. 6. G6ncz61 E, Ianacone J, Ho W, Starr S, Meignier B, Plotkin S. Isolated gA/gB glycoprotein complex of human cytomegalovirus envelope induces humoral and cellular immune-responses in human volunteers. Vaccine 1990;8: 130-6. 7. Gold 0, Ashley R, Handsfield HH, et al. Immunoblot analysis of the humoral immune response in primary cytomegalovirus infection. J Infect Dis 1988;157:319-26. 8. Liu YNC Kari B, Gehrz RC Human immune responses to major human cytomegalovirus glycoprotein complexes. J Virol 1988; 62: 1066-70. 9. Britt WJ, Vugler L, Butfiloski EJ, Stephens EB. Cell surface expression of human cytomegalovirus (HCMV) gp55-116 (gB): use of HCMVrecombinant vaccinia virus-infected cells in analysis of the human neutralizing antibody response. J Virol 1990;64: 1079-85. 10. Spaete RR, Thayer RM, Probert WS, et al. Human cytomegalovirus strain Towne glycoprotein B is processed by proteolytic cleavage. Virology 1988;167:207-25. II. Chou S, Dennison KM. Analysis of interstrain variation in cytomegalovirus glycoprotein B sequences encoding neutralization-related epitopes. J Infect Dis 1991; 163: 1229-34. 12. Britt WJ, Vugler LG. Processing of the gp55-116 envelope glycoprotein complex of human cytomegalovirus. J Viral 1989;63:403-10. 13. Spaete RR, Saxena A, Scott PI, et al. Sequence requirements for proteolytic processing of glycoprotein B of human cytomegalovirus strain Towne. J Virol 1990;64:2922-31. 14. Kniess N, Mach M, Fay J, Britt WJ. Distribution of linear antigenic sites on glycoprotein gp55 of human cytomegalovirus. J Virol 1991;65: 138-46. 15. Meyer H, Masuho Y, Mach M. The gpl16 of the gp58/116 complex of human cytomegalovirus represents the amino-terminal part of the precursor molecule and contains a neutralizing epitope. J Gen Virol 1990;71 :2443-50.

Downloaded from http://jid.oxfordjournals.org/ at University of Iowa Libraries/Serials Acquisitions on June 25, 2015

a context isolated from other HCMV proteins. gB antibody was found in all sera that had antibody to HCMV. This study extends earlier work using physically purified antigen [8] by demonstrating a strong correlation between serum neutralizing activity and gB antibody levels in the absence of a strong correlation between total HCMVspecific IgG and gB antibody. While it is difficult to establish a causal relationship on the basis of these correlations alone, the conclusion that a significant fraction ofneutralizing activity comes directly from gB antibody is supported by specific absorption experiments. Absorption of20 sera with an insoluble antigen preparation resulted in the median loss of 48% of neutralizing activity. There was a wide range in reduction of neutralizing activity, so that for some sera almost 100% of activity was removed by absorption, whereas for others almost all neutralizing activity was retained. Although it is possible that not all neutralizing epitopes were presented on the absorbent, similar results were reported using intact cells infected with a vaccinia-gB recombinant [9]; in that study, 26%-100% of neutralizing activity was removed from 12 convalescent sera. Since other HCMV envelope glycoproteins are known to carry neutralizing epitopes, it is possible that some of the residual activity in these absorption experiments is due to antibody to these antigens. Antigenic variability in gB between infecting strains could account for some of the differences in neutralizing titer between subjects in this study, but a recent sequence analysis suggests that prominent neutralizing epitopes appear to be well conserved among strains [ 11 ]. In HCMV-infected cells, gB initially appears in fully glycosylated forms of 130-170 kDa from which the 58- to 52kDa carboxy terminus is subsequently cleaved [10, 12, 13]. This fragment, termed gp55, is the immunodominant part of the mature protein [14]. The amino terminal cleavage product, which is referred to as gp93 by some authors [10] and gp 116 by others [12], is less abundant in virions, but it does carry neutralizing epitopes and it is immunogenic [15]. In the current study, specific bands in the range of 93-116 kDa were not detected by RIP in Ad-gB-infected cell antigens with any antisera. As this protein is heavily glycosylated and is associated with gp55 through disulfide bonds, it should copurii)' with gp55 on lectin columns and should coprecipitate in RIP assays. Failure to detect this protein might indicate that the processing of gB is different in adenovirus-infected cells than in HCMV-infected fibroblasts, as suggested in an earlier study [5]. Cranage et al. [4] were also unable to detect the amino-terminal fragment in cells infected with a vaccinia-gB recombinant and suggested that the protein might be rapidly degraded. In contrast, Britt et al. [9] have demonstrated surface expression of gp 116 in HeLa cells infected with a similar recombinant vaccinia virus. The detailed processing of gB in Ad-gB-infected cells is currently under investigation. Although cellular immune responses are probably more important in controlling established HCMV infection, there

110 1992; 165 (February)